Skip to main content

​
ChatHistoryMemory

class ChatHistoryMemory(AgentMemory):
An agent memory wrapper of :obj:ChatHistoryBlock. Parameters:
  • context_creator (BaseContextCreator): A model context creator.
  • storage (BaseKeyValueStorage, optional): A storage backend for storing chat history. If None, an :obj:InMemoryKeyValueStorage will be used. (default: :obj:None)
  • window_size (int, optional): The number of recent chat messages to retrieve. If not provided, the entire chat history will be retrieved. (default: :obj:None)
  • agent_id (str, optional): The ID of the agent associated with the chat history.

​
init

def __init__(
    self,
    context_creator: BaseContextCreator,
    storage: Optional[BaseKeyValueStorage] = None,
    window_size: Optional[int] = None,
    agent_id: Optional[str] = None
):

​
agent_id

def agent_id(self):

​
agent_id

def agent_id(self, val: Optional[str]):

​
retrieve

def retrieve(self):

​
write_records

def write_records(self, records: List[MemoryRecord]):

​
get_context_creator

def get_context_creator(self):

​
clear

def clear(self):

​
clean_tool_calls

def clean_tool_calls(self):
Removes tool call messages from memory. This method removes all FUNCTION/TOOL role messages and any ASSISTANT messages that contain tool_calls in their meta_dict to save token usage.

​
VectorDBMemory

class VectorDBMemory(AgentMemory):
An agent memory wrapper of :obj:VectorDBBlock. This memory queries messages stored in the vector database. Notice that the most recent messages will not be added to the context. Parameters:
  • context_creator (BaseContextCreator): A model context creator.
  • storage (BaseVectorStorage, optional): A vector storage storage. If None, an :obj:QdrantStorage will be used. (default: :obj:None)
  • retrieve_limit (int, optional): The maximum number of messages to be added into the context. (default: :obj:3)
  • agent_id (str, optional): The ID of the agent associated with the messages stored in the vector database.

​
init

def __init__(
    self,
    context_creator: BaseContextCreator,
    storage: Optional[BaseVectorStorage] = None,
    retrieve_limit: int = 3,
    agent_id: Optional[str] = None
):

​
agent_id

def agent_id(self):

​
agent_id

def agent_id(self, val: Optional[str]):

​
retrieve

def retrieve(self):

​
write_records

def write_records(self, records: List[MemoryRecord]):

​
get_context_creator

def get_context_creator(self):

​
clear

def clear(self):
Removes all records from the vector database memory.

​
LongtermAgentMemory

class LongtermAgentMemory(AgentMemory):
An implementation of the :obj:AgentMemory abstract base class for augmenting ChatHistoryMemory with VectorDBMemory. Parameters:
  • context_creator (BaseContextCreator): A model context creator.
  • chat_history_block (Optional[ChatHistoryBlock], optional): A chat history block. If None, a :obj:ChatHistoryBlock will be used. (default: :obj:None)
  • vector_db_block (Optional[VectorDBBlock], optional): A vector database block. If None, a :obj:VectorDBBlock will be used. (default: :obj:None)
  • retrieve_limit (int, optional): The maximum number of messages to be added into the context. (default: :obj:3)
  • agent_id (str, optional): The ID of the agent associated with the chat history and the messages stored in the vector database.

​
init

def __init__(
    self,
    context_creator: BaseContextCreator,
    chat_history_block: Optional[ChatHistoryBlock] = None,
    vector_db_block: Optional[VectorDBBlock] = None,
    retrieve_limit: int = 3,
    agent_id: Optional[str] = None
):

​
agent_id

def agent_id(self):

​
agent_id

def agent_id(self, val: Optional[str]):

​
get_context_creator

def get_context_creator(self):
Returns: BaseContextCreator: The context creator used by the memory.

​
retrieve

def retrieve(self):
Returns: List[ContextRecord]: A list of context records retrieved from both the chat history and the vector database.

​
write_records

def write_records(self, records: List[MemoryRecord]):
Converts the provided chat messages into vector representations and writes them to the vector database. Parameters:
  • records (List[MemoryRecord]): Messages to be added to the vector database.

​
clear

def clear(self):
Removes all records from the memory.
⌘I